Search

Autonomous vehicle advancements and communication technologies such as V2V, V2I, and V2X have enabled the development of connected and autonomous vehicles. Because CAVs are directly effective in traffic, their application in traffic management and incident management appears promising. They can immediately begin regulating traffic and acting as sensors due to their connectivity to the infrastructure. This research proposes Incident Detection Included Linear Management (IDILIM), a CAV-based incident management algorithm that regulates CAV and traffic speeds based on dynamic and predicted shockwave speeds. The SUMO simulations are carried out on a 10.4-km-long, three-lane facility with 21 sensors every 500 m. In the scenarios, three traffic demands, eleven CAV penetration rates, and varying incident locations, duration, and lanes are used. A total of 20 simulation seeds are used in each scenario. The proposed algorithm necessitates the use of a reliable traffic prediction model. Convolutional Neural Networks, a deep learning algorithm with high estimation accuracy, are used in the prediction model. IDILIM uses the highly accurate traffic prediction output of the Pix-to-Pix model as input at 3-min intervals. Shockwave speed is calculated using model outputs and fed to CAVs. To compare with IDILIM, variable speed limits (VSL) are also modeled. When compared to uncontrolled base scenarios, IDILIM reduced density values greater than 35 veh/km in the critical region by 89.32%. In the same scenario, VSL management decreased by only 52.43%. [ABSTRACT FROM AUTHOR]

Within the scope of this paper we examine incident detection on inner-city roads using floating car data (FCD) under different traffic conditions and data qualities. We used synthetic FCD to analyze incidents under variable conditions, such as different traffic volumes. They are generated from microscopic traffic flow simulations with the software tool PTV Vissim. Automated processing of the synthetic FCD gained from the microsimulation models also makes it possible to analyze different data qualities resulting from the choice of penetration rate and transmission frequency. We investigated the recognizability of the incidents using five laboratory examples, which depict different traffic incidents, and a real inner-city road in Karlsruhe, Germany. Graphical and quantitative evaluations are used for incident detection, whereby violin plots and the proposed incident indicator have the most reliable detection rate. Incident detection decreases significantly if several incidents overlap, the severity of the incidents or the traffic volume is reduced, the data quality decreases, or if there is no current speed information in the data. For additional insights, we recommend considering the different figures created in the graphical analysis alongside the incident indicator used in the quantitative analysis. [ABSTRACT FROM AUTHOR]

In the realm of modern urban mobility, automatic incident detection is a critical element of intelligent transportation systems (ITS), since the ability to promptly identify unexpected events allows for quick implementation of preventive measures and efficient response to the situations as they arise. With the growing availability of traffic data, Machine Learning (ML) has become a vital tool for enhancing traditional incident detection methods. Automated machine-learning (AutoML) techniques present a promising solution by streamlining the machine-learning process; however the application of AutoML for incident detection has not been widely explored in scientific research In this paper, we propose and apply an AutoML-based methodology for traffic incident detection and compare it with state-ofthe-art ML approaches. Our approach integrates data preprocessing with AutoML, and uses Tree-based Pipeline Optimization Tool (TPOT) to refine the process from raw data to prediction. We have tested the efficiency of our approach in two major European cities, Athens and Antwerp. Finally, we present the limitations of our work and outline recommendations for application of AutoML in the incident detection task and potentially in other domains. [ABSTRACT FROM AUTHOR]

Accidents can have a significant impact on road safety and the efficiency of transport. With the ever-increasing number of vehicles on the road, it is crucial to quickly detect and respond to such events to decrease the impact. The current paper presents a road accident and incident detection solution utilizing traffic API monitoring and continuously analyzing real-time traffic data from various sources like GPS-enabled devices. Consequently, the proposed approach can identify abnormal patterns in the movement of vehicles, such as unexpected road closures or an accident. It numerically analyzes the traffic pattern and identifies incidents and accidents using live traffic data collected from the map service. When the approach identifies any abnormality, it can immediately let the user know, allowing them to act accordingly. It can facilitate authorities in identifying accident-prone areas and deploy safety measures. This can help law enforcement agencies to identify drivers at risk of causing accidents and implement education and enforcement programs targeted towards them. Hence, the proposed approach can aid in reducing the negative impact of traffic incidents, and the public can respond to these incidents more preparedly. [ABSTRACT FROM AUTHOR]

Time of incident reporting is a critical aspect of emergency response. However, the conventional approaches to receiving incident reports have time delays. Non-traditional sources such as crowdsourced data present an opportunity to detect incidents proactively. However, detecting incidents from such data streams is challenging due to inherent noise and data uncertainty. Naively maximizing detection accuracy can compromise spatial-temporal localization of inferred incidents, hindering response efforts. This article presents a novel human-centered AI tool to address the above challenges. We demonstrate how crowdsourced data can aid incident detection while acknowledging associated challenges. We use an existing CROME framework to facilitate training and selection of best incident detection models, based on parameters suited for deployment. The human-centered AI tool provides a visual interface for exploring various measures to analyze the models for the practitioner's needs, which could help the practitioners select the best model for their situation. Moreover, in this study, we illustrate the tool usage by comparing different models for incident detection. The experiments demonstrate that the CNN-based incident detection method can detect incidents significantly better than various alternative modeling approaches. In summary, this research demonstrates a promising application of human-centered AI tools for incident detection to support emergency response agencies. [ABSTRACT FROM AUTHOR]

Cooperative vehicular systems require that vehicles fuse sensor data and broadcast one-hop safety messages containing their kinematic information to enable vehicular applications based on incident detection. Several congestion control mechanisms have been proposed to mitigate channel congestion resulting from the continuous transmission of safety messages. This paper investigates the effect of message rate adaptation-based congestion control from a road safety perspective by evaluating the feasibility of prominent approaches, such as PULSAR, LIMERIC, reactive ETSI DCC, and SAE J2945/1, to support lane-changing maneuvers on multi-lane highways under varying conditions. Simulation results demonstrate that message size, vehicular density, losses at the physical layer, and observation time significantly influence the lane-changing application’s capability to detect unsafe maneuvers when congestion control is in action. Specific recommendations and guidelines for congestion control are provided to improve decision-making at the application level.

The task of designing and training a 3D convolutional neural network (CNN) from scratch poses significant complexity, necessitating high levels of expertise to achieve a performance that rivals the state-of-the-art. To circumvent this, fine-tuning of neural networks has emerged as a formidable approach. This study focuses on the utilization of Video ResNet, a state-of-the-art architecture known for its proficiency in capturing spatiotemporal patterns from video data. A novel approach is proposed for the fine-tuning of the 3D CNN model (Video ResNet) that involves altering activation functions over epochs while maintaining the network weights and biases consistent. This dynamic approach was assessed under various hyperparameters, yielding encouraging results. Contrary to most studies that employ down-sampling of the temporal sequence to minimize memory requirements, this study introduces a sliding window-based approach to evade down-sampling and prevent potential information loss. The proposed methodology yielded an accuracy of 87.25% in the fight/non-fight classification on the RWF-2000 dataset, marginally surpassing the performance of the state-of-the-art model. The proposed method not only facilitates the development of a real-time video incident detection model but also addresses the issue of overfitting during training through the incorporation of adaptive dynamic activation functions. This study thus contributes to the ongoing advancements in the field of neural network fine-tuning and video data classification. [ABSTRACT FROM AUTHOR]

Information shared on social media by transit system customers is often candid, localized, and includes in the moment information about emerging events or issues. Twitter provides an unfiltered and timestamped feed of information that can be aggregated to generate valuable insights. Our research aims to identify passenger-related transit incidents from a public Twitter feed. Detecting these incidents in real time enables transit agencies to immediately respond to them by dispatching security, safety, or maintenance crews or by rapidly replying to requests made to the agency that are urgent in nature. We leverage natural language processing to extract latent information from identified eyewitness tweets about transit, focusing on location details, topic classification, and sentiment analysis. We outline an approach to developing a useful corpus of transit-focused tweets, detecting in the moment events, classifying these tweets into topics, and detecting locations where possible. We then demonstrate the approach through an application to Calgary Transit in Calgary, Canada. The results demonstrate that key incidents can be identified and prioritized for an agency. [ABSTRACT FROM AUTHOR]

Machine Learning (ML) represents a pivotal technology for current and future information systems, and many domains already leverage the capabilities of ML. However, deployment of ML in cybersecurity is still at an early stage, revealing a significant discrepancy between research and practice. Such a discrepancy has its root cause in the current state of the art, which does not allow us to identify the role of ML in cybersecurity. The full potential of ML will never be unleashed unless its pros and cons are understood by a broad audience. This article is the first attempt to provide a holistic understanding of the role of ML in the entire cybersecurity domain—to any potential reader with an interest in this topic. We highlight the advantages of ML with respect to human-driven detection methods, as well as the additional tasks that can be addressed by ML in cybersecurity. Moreover, we elucidate various intrinsic problems affecting real ML deployments in cybersecurity. Finally, we present how various stakeholders can contribute to future developments of ML in cybersecurity, which is essential for further progress in this field. Our contributions are complemented with two real case studies describing industrial applications of ML as defense against cyber-threats. [ABSTRACT FROM AUTHOR]

This study presents a comprehensive framework for optimizing intelligent transport systems (ITS) by integrating advanced communication and information technologies into vehicles, roads, and infrastructure. The primary goal is to enhance transportation efficiency, safety, and environmental sustainability while improving overall mobility for people and goods. Leveraging contextual information, the framework offers personalized, proactive services such as real-time traffic updates, route recommendations, and parking availability. Additionally, it enhances safety and security by providing early hazard warnings and adapting to changing road conditions. Our proposed framework utilizes the enhanced coral reef optimization (ECRO) algorithm to efficiently group vehicles for energy-saving data collection, maximizing information gathering efficiency. Collected data is then transmitted to a central data gathering center via a sink node optimized through the modified pelican optimization (MPO) algorithm, considering various vehicle node design constraints. An incident detection module accurately classifies and detects road incidents, enabling timely emergency service requests and alternate route recommendations. To facilitate incident detection, we introduce the deep Rigdelet neural network (DRNN), a novel deep learning technique tailored for decision-making in incident classification. We validate our framework's performance through NS-2 simulations using the SUMO traffic generator, demonstrating its effectiveness in meeting quality of service (QoS) metrics. Through comparative analysis with existing frameworks, our proposed approach stands out for its superior performance and ability to optimize ITS operations. [ABSTRACT FROM AUTHOR]

Multivariate time series classification is a machine learning problem that can be applied to automate a wide range of real-world data analysis tasks. RandOm Convolutional KErnel Transform (ROCKET) proved to be an outstanding algorithm capable to classify time series accurately and quickly. The textbook variant of the multivariate time series classification problem assumes that time series to be classified are all of the same length, while in real-world applications this assumption not necessarily holds. The literature of this domain does not pay enough attention to data processing pipelines for variable-length time series. Thus, in this paper, we present a thorough analysis of three preprocessing pipelines that handle variable-length time series that need to be classified with a method that requires the data to be of equal length. These three methods are truncation, padding, and forecasting of missing value. Experiments conducted on benchmark datasets, showed that the recommended procedure involves padding. Forecasting ensures similar classification accuracy, but comes at a much higher computational cost. Truncation is not a viable option. Furthermore, in the paper, we present a novel domain of application of multivariate time series classification algorithms, that is incident detection in cash transactions. This area poses substantive challenges for automated model training procedures since the data is not only variable-length, but also heavily imbalanced. In the study, we list various incident types and present trained classifiers capable to aid human auditors in their daily work.

Non-recurrent congestion disrupts normal traffic operations and lowers travel time (TT) reliability, which leads to many negative consequences such as difficulties in trip planning, missed appointments, loss in productivity, and driver frustration. Traffic incidents are one of the six causes of non-recurrent congestion. Early and accurate detection helps reduce incident duration, but it remains a challenge due to the limitation of current sensor technologies. In this paper, we employ a recurrence-based technique, the Quadrant Scan, to analyse time series traffic volume data for incident detection. The data is recorded by multiple sensors along a section of urban highway. The results show that the proposed method can detect incidents better by integrating data from the multiple sensors in each direction, compared to using them individually. It can also distinguish non-recurrent traffic congestion caused by incidents from recurrent congestion. The results show that the Quadrant Scan is a promising algorithm for real-time traffic incident detection with a short delay. It could also be extended to other non-recurrent congestion types. [ABSTRACT FROM AUTHOR]

Network Security Monitoring (NSM) is a popular term to refer to the detection of security incidents by monitoring the network events. An NSM system is central for the security of current networks, given the escalation in sophistication of cyberwarfare. In this paper, we review the state-of-the-art in NSM, and derive a new taxonomy of the functionalities and modules in an NSM system. This taxonomy is useful to assess current NSM deployments and tools for both researchers and practitioners. We organize a list of popular tools according to this new taxonomy, and identify challenges in the application of NSM in modern network deployments, like Software Defined Network (SDN) and Internet of Things (IoT).

The surge of renewable energy systems can lead to increasing incidents that negatively impact economics and society, rendering incident detection paramount to understand the mechanism and range of those impacts. In this paper, a deep learning framework is proposed to detect renewable energy incidents from news articles containing accidents in various renewable energy systems. The pre-trained language models like Bidirectional Encoder Representations from Transformers (BERT) and word2vec are utilized to represent textual inputs, which are trained by the Text Convolutional Neural Networks (TCNNs) and Text Recurrent Neural Networks. Two types of classifiers for incident detection are trained and tested in this paper, one is a binary classifier for detecting the existence of an incident, the other is a multi-label classifier for identifying different incident attributes such as causal-effects and consequences, etc. The proposed incident detection framework is implemented on a hand-annotated dataset with 5 190 records. The results show that the proposed framework performs well on both the incident existence detection task (F1-score 91.4%) and the incident attributes identification task (micro F1-score 81.7%). It is also shown that the BERT-based TCNNs are effective and robust in detecting renewable energy incidents from large-scale textual materials. [ABSTRACT FROM AUTHOR]

The timely and accurate detection of traffic incidents is beneficial to reduce associated economic losses and avoid secondary crashes. Inspired by the impressive success of the image classification algorithms, especially convolutional neural networks (CNNs), this study proposes a novel framework to detect highway traffic incidents by learning the traffic state as images. In such a framework, the probe vehicles equipped with the global positioning system equipment are used to obtain data. The Gramian Angular Difference Fields and Piecewise Aggregation Approximation algorithms are used to convert the link speed time series data into images. CNNs can extract the traffic features based on these images and consider an incident detection problem as a binary classification task. Further, the effectiveness of the proposed framework is evaluated by applying it to detect the traffic in a real-world environment, i.e., the Guangzhou Airport Expressway. The results illustrate that the proposed model outperforms several other algorithms with respect to almost all the performance indexes, including the detection rate with different false alarm rates and the area under the receiver operating characteristic curve.

As a component of intelligent transport systems (ITS), vehicular ad hoc network (VANET), which is a subform of manet, has been identified. It is established on the roads based on available vehicles and supporting road infrastructure, such as base stations. An accident can be defined as any activity in the environment that may be harmful to human life or dangerous to human life. In terms of early detection, and broadcast delay. VANET has shown various problems. The available technologies for incident detection and the corresponding algorithms for processing. The present problem and challenges of incident detection in VANET technology are discussed in this paper. The paper also reviews the recently proposed methods for early incident techniques and studies them. [ABSTRACT FROM AUTHOR]

With a large of time series dataset from the Internet of Things in Ambient Intelligence-enabled smart environments, many supervised learning-based anomaly detection methods have been investigated but ignored the correlation among the time series. To address this issue, we present a new idea for anomaly detection based on dynamic graph embedding, in which the dynamic graph comprises the multiple time series and their correlation in each time interval. We propose an entropy for measuring a graph's information injunction with a correlation matrix to define similarity between graphs. A dynamic graph embedding model based on the graph similarity is proposed to cluster the graphs for anomaly detection. We implement the proposed model in vehicular edge computing for traffic incident detection. The experiments are carried out using traffic data produced by the Simulation of Urban Mobility framework. The experimental findings reveal that the proposed method achieves better results than the baselines by 14.5% and 18.1% on average with respect to F1-score and accuracy, respectively. [ABSTRACT FROM AUTHOR]

Geological drilling process is operating under complex geological conditions, which may lead to a high risk of downhole incidents and thus compromise the drilling efficiency. To achieve prompt detection of downhole incidents and prevent them from developing to serious drilling accidents, this paper proposes a new data-driven detection method for downhole incidents based on the amplitude change detection and dynamic time warping. Two major phases are involved: the change monitoring phase detects whether there is any significant change in the drilling signals and extracts variational trend features by linear fitting and amplitude change detection; the incident detection phase determines if the cause of a change is a normal switching or a downhole incident by similarity analysis based on the dynamic time warping and the density-based spatial clustering. Industrial case studies show that the proposed method achieves good performance in downhole incidents detection for geological drilling processes. • This work proposes a new data-driven method to detect downhole incidents. • A change monitoring approach is presented to detect signal amplitude changes. • A similarity analysis technique is developed based on DTW and DBSCAN. [ABSTRACT FROM AUTHOR]

The adequacy of traditional transport related issues detection is often limited by physical sparse sensor coverage and reporting incident/issues to the emergency response system is labor intensive. The social media tweet text have been mined so as to identify the complaints regarding various road transportation issues of traffic, accident, and potholes. In order to identify and segregate tweets related to different issues, keyword-based approaches have been used previously, but these methods are solely dependent on seed keywords which are manually given and these set of keywords are not sufficient to cover all tweets posts. So, to overcome this issue, a novel approach has been proposed that captures the semantic context through dense word embedding by employing word2vec model. However, the process of tweet segregation on the basis of semantic similar keywords may suffer from the problem of pragmatic ambiguity. To handle this, Word2Vec model has been applied to match the semantically similar tweets with respect to each category. Furthermore, the hotspots have been identified corresponding to each category. However, due to the scarcity of geo-tagged tweets, we have proposed a hybrid method which amalgamates Named Entity Recognition (NER), Part of speech (POS), and Regular Expression (RE) to extract the location information from the tweet textual content. Due to the lack of availability of the ground truth dataset, model feasibility has been validated from the existing data records (i.e., published by government official accounts and reported on news media) and the evaluation results signify that the stated approach identifies few additional hotspots as compared to the existing reports while analyzing the tweets.

In autonomous driving, IEEE 802.11p-based vehicle-to-vehicle (V2V) communication is considered for overcoming the intrinsic limitations of sensors and supporting safety applications. In this letter, we evaluate the effectiveness of relevant awareness control approaches, such as ETSI DMG, IVTRC, and POSACC, to support the V2V-based overtaking application in autonomous driving. For this, we assess the incident detection capability of the overtaking application when it is running with messages gathered from these approaches, considering packet losses due to channel fading. Simulations show that POSACC is more effective than the remaining approaches for detecting unsafe overtaking maneuvers in different operating conditions. [ABSTRACT FROM AUTHOR]

Non-recurrent congestion disrupts normal traffic operations and lowers travel time (TT) reliability, which leads to many negative consequences such as difficulties in trip planning, missed appointments, loss in productivity, and driver frustration. Traffic incidents are one of the six causes of non-recurrent congestion. Early and accurate detection helps reduce incident duration, but it remains a challenge due to the limitation of current sensor technologies. In this paper, we employ a recurrence-based technique, the Quadrant Scan, to analyse time series traffic volume data for incident detection. The data is recorded by multiple sensors along a section of urban highway. The results show that the proposed method can detect incidents better by integrating data from the multiple sensors in each direction, compared to using them individually. It can also distinguish non-recurrent traffic congestion caused by incidents from recurrent congestion. The results show that the Quadrant Scan is a promising algorithm for real-time traffic incident detection with a short delay. It could also be extended to other non-recurrent congestion types.

In this work, we propose the Road Data Enrichment (RoDE), a framework that fuses data from heterogeneous data sources to enhance Intelligent Transportation System (ITS) services, such as vehicle routing and traffic event detection. We describe RoDE through two services: (i) Route service, and (ii) Event service. For the first service, we present the Twitter MAPS (T-MAPS), a low-cost spatiotemporal model to improve the description of traffic conditions through Location-Based Social Media (LBSM) data. As a case study, we explain how T-MAPS is able to enhance routing and trajectory descriptions by using tweets. Our experiments compare T-MAPS’ routes against Google Maps’ routes, showing up to 62% of route similarity, even though T-MAPS uses fewer and coarse-grained data. We then propose three applications, Route Sentiment (RS), Route Information (RI), and Area Tags (AT), to enrich T-MAPS’ suggested routes. For the second service, we present the Twitter Incident (T-Incident), a low-cost learning-based road incident detection and enrichment approach built using heterogeneous data fusion. Our approach uses a learning-based model to identify patterns on social media data which is then used to describe a class of events, aiming to detect different types of events. Our model to detect events achieved scores above 90%, thus allowing incident detection and description as a RoDE application. As a result, the enriched event description allows ITS to better understand the LBSM user’s viewpoint about traffic events (e.g., jams) and points of interest (e.g., restaurants, theaters, stadiums). [ABSTRACT FROM AUTHOR]

This study proposes formulation of a system for incident detection and management of traffic signals constituting urban traffic networks. A system prototype has been developed and tested in a simulation environment under several incident scenarios. Following incident detection, the proposed system deploys a multistage fuzzy-logic model (FLM) to manage traffic signals. Details of FLM calibration have been presented and discussed. The proposed system has been calibrated under various traffic conditions and incident scenarios. A parametric sensitivity analysis was performed to optimize the proposed FLM, and further analysis has been performed to demonstrate robustness when tested under conditions different from those it has been optimized for, thereby leading to development of response surface methodology (RSM) models to determine the most robust parameters of FLM. RSM has been calibrated using the Box–Behnken design (BBD). Three different non-linear regression models have been used to identify those robust parameters concerning incident detection and traffic management that are likely to minimize the overall network travel time. [ABSTRACT FROM AUTHOR]

This study proposes a statistical approach to incident detection in a section of the intercity freeway by applying GPS probe data to a GIS geofenced platform. We evaluated the proposed method using data sources from real traffic sensors of the intercity Tehran-Qom freeway in Iran. Through the SEPEHTAN project in Iran, intercity bus fleet equipped with an onboard unit that provides GPS data transferring to the central database. The main novelties in this paper are gathering density and speed time series from GPS probe data in a GIS platform and using autocorrelation factor to detect the location of the incident. The method compared with three different AID algorithms and real terms as well. Although the penetration rate was 3%, the results were considerably meet with the actual traffic condition. We reached 92.8% detection rate and 7.1% for the false alarm. [ABSTRACT FROM AUTHOR]

Freeway accidents cause grave fatalities and economic losses. Further losses of life and property can be caused by secondary accidents or rear-end chain accidents. Therefore, a quick and accurate incident detection technique is vital to improve the road safety and traffic management system performance. Based on the set theory and data collected by smartphones, a low-cost incident detection algorithm is proposed, called set theory-based freeway incident detection algorithm (SFID). If a vehicle is involved in a traffic incident, the capacity of the road segment in one direction declines significantly, whereas there are no changes in the opposite direction. SFID uses this phenomenon to infer if an incident has occurred or not. To evaluate the performance of SFID, simulations are conducted under different conditions. The results show a high detection rate, low average detection time, and low false alarm rate. Furthermore, the proposed scheme is evaluated using real Wi-Fi data. The results are in concordance with those of simulations. [ABSTRACT FROM AUTHOR]

With a large of time series dataset from the Internet of Things in Ambient Intelligence-enabled smart environments, many supervised learning-based anomaly detection methods have been investigated but ignored the correlation among the time series. To address this issue, we present a new idea for anomaly detection based on dynamic graph embedding, in which the dynamic graph comprises the multiple time series and their correlation in each time interval. We propose an entropy for measuring a graph’s information injunction with a correlation matrix to define similarity between graphs. A dynamic graph embedding model based on the graph similarity is proposed to cluster the graphs for anomaly detection. We implement the proposed model in vehicular edge computing for traffic incident detection. The experiments are carried out using traffic data produced by the Simulation of Urban Mobility framework. The experimental findings reveal that the proposed method achieves better results than the baselines by 14.5% and 18.1% on average with respect to F1-score and accuracy, respectively.

Prior research on traffic event detection has encountered two problems: limited sample numbers and unbalanced datasets. Moreover, the real-time properties of event detection models must be enhanced to meet traffic management demands. To solve these issues, suitable measures must be implemented, like developing an intelligent algorithm for incident detection employing Artificial Intelligence (AI) and Machine Learning (ML). Automated Incident Detection (AID) methods are the focus of current Intelligent Transportation System (ITS) technology. Modern vehicles can connect with one other and with Roadside Infrastructure Units (RSUs) to improve road safety thanks to advancements in wireless connectivity and sensor technology. Deep Learning (DL)-based methods have recently demonstrated strong performance in computer vision issues involving complicated feature associations. This work proposes a Hybrid Deep Learning-based Automated Incident Detection and Management (HDL-AIDM) system to identify traffic incidents and improve traffic management. In the suggested model, a Temporal and Spatial Stacked Autoencoder (TSSAE) is used to collect temporal and spatial associations of traffic conditions and identify events. At the same time, a generative adversarial network (GAN) is employed to improve the number of samples and equalize datasets. The proposed model is assessed from many perspectives using the dataset from real-world situations. Based on the HDL output for AID, an efficient and intelligent traffic management algorithm has been accomplished using Road Side Units (RSUs) to gather traffic information. The suggested incident management algorithm considers lane shifts and the fluctuation in vehicle speed over time, which are heavily influenced by traffic incidents. These developments in ITS enable traffic management systems to use information gathered from HDL-based AID methods using TSSAE and GAN. The proposed strategies have been devised to notify drivers about traffic issues and help them avoid congestion. The suggested method provides higher incident detection rates with an accuracy of 94.1%, a 3.9% false alarm rate, and an incident classification rate of 93.3%. [ABSTRACT FROM AUTHOR]

This research paper investigates a hybrid model using logistic regression with a wavelet-based feature extraction for detecting traffic incidents. A logistic regression model is suitable when the outcome can take only a limited number of values. For traffic incident detection, the outcome is limited to only two values, the presence or absence of an incident. The logistic regression model used in this study is a generalized linear model (GLM) with a binomial response and a logit link function. This paper presents a framework to use logistic regression and wavelet-based feature extraction for traffic incident detection. It investigates the effect of preprocessing data on the performance of incident detection models. Results of this study indicate that logistic regression along with wavelet based feature extraction can be used effectively for incident detection by balancing the incident detection rate and the false alarm rate according to need. Logistic regression on raw data resulted in a maximum detection rate of 95.4% at the cost of 14.5% false alarm rate. Whereas the hybrid model achieved a maximum detection rate of 98.78% at the expense of 6.5% false alarm rate. Results indicate that the proposed approach is practical and efficient; with future improvements in the proposed technique, it will make an effective tool for traffic incident detection.

In this paper, the authors describe the successful combination of a low- level, vision-based surveillance system with a high-level, symbolic reasoner based on dynamic belief networks. This prototype system provides robust, high-level information about traffic scenes. The machine vision component of the system employs a correlation-based tracker and a physical motion model using a Kalman filter to extract vehicle trajectories over a sequence of traffic scene images. The symbolic reasoning component uses a dynamic belief network to make inferences about traffic events. In this paper, the authors discuss the key tasks of the vision and reasoning components as well as their integration into a working prototype.

We introduce a novel methodology for anomaly detection in time-series data. The method uses persistence diagrams and bottleneck distances to identify anomalies. Specifically, we generate multiple predictors by randomly bagging the data (reference bags), then for each data point replacing the data point for a randomly chosen point in each bag (modified bags). The predictors then are the set of bottleneck distances for the reference/modified bag pairs. We prove the stability of the predictors as the number of bags increases. We apply our methodology to traffic data and measure the performance for identifying known incidents.